Contrasting Effects of Elevated Temperature and Invertebrate Grazing Regulate Multispecies Interactions between Decomposer Fungi

Predicting the influence of biotic and abiotic factors on species interactions and ecosystem processes is among the primary aims of community ecologists. The composition of saprotrophic fungal communities is a consequence of competitive mycelial interactions, and a major determinant of woodland decomposition and nutrient cycling rates. Elevation of atmospheric temperature is predicted to drive changes in fungal community development. Top-down regulation of mycelial growth is an important determinant of, and moderator of temperature-driven changes to, two-species interaction outcomes. This study explores the interactive effects of a 4 °C temperature increase and soil invertebrate (collembola or woodlice) grazing on multispecies interactions between cord-forming basidiomycete fungi emerging from colonised beech (Fagus sylvatica) wood blocks. The fungal dominance hierarchy at ambient temperature (16 °C; Phanerochaete velutina > Resinicium bicolor > Hypholoma fasciculare) was altered by elevated temperature (20 °C; R. bicolor > P. velutina > H. fasciculare) in ungrazed systems. Warming promoted the competitive ability of the fungal species (R. bicolor) that was preferentially grazed by all invertebrate species. As a consequence, grazing prevented the effect of temperature on fungal community development and maintained a multispecies assemblage. Decomposition of fungal-colonised wood was stimulated by warming, with implications for increased CO2 efflux from woodland soil. Analogous to aboveground plant communities, increasing complexity of biotic and abiotic interactions appears to be important in buffering climate change effects on soil decomposers.


Introduction
The influence of environmental change on species interactions is a major determinant of community structure and function [1]. Predicting how biotic and abiotic context affects the responses of species interactions and ecosystem processes to climate change remains one of the greatest challenges in community ecology [2]. Though the key determinants of plant community dynamics have been extensively explored (e.g. [3][4][5]), little is known about belowground community responses to climate change. While the importance of soil microbial communities in carbon and nutrient cycling is widely acknowledged [6,7], the mechanisms regulating their composition and functioning remain poorly resolved. Fungi, in particular, exert a strong influence on soil organic matter dynamics and the temperature-sensitivity of decomposition [8]. The production of lignocellulolytic enzymes by saprotrophic basidiomycete fungi makes them primary regulators of decomposition in temperate woodland ecosystems [9,10]. Filamentous networks of mycelium formed by a major ecological grouping of these fungi, the saprotrophic cord-formers, ramify at the soil-litter interface where they forage for woody resources [11,12]. Competitive mycelial interactions determine fungal dominance and community composition [13]. These fungi largely compete, for space first and foremost, by antagonistic mechanisms commonly referred to as combat [13]. Factors affecting competitive fungal interaction outcomes will influence nutrient cycling and CO 2 efflux from soil, as a consequence of species-specific extracellular enzyme production and resource decomposition rates [14,15].
Temperature is an important abiotic determinant of decomposer fungal growth and activity [16]. Differential warminginduced stimulation of competing saprotrophic basidiomycete mycelial growth and combative ability has been shown to alter interaction outcomes in soil [17,18]. During mycelial interactions, elevated temperature only increases the growth of the dominant competitor; increased growth by the inferior competitors is prevented by antagonism (overgrowth or direct killing of mycelia) during combative interactions [17]. Warming may also indirectly influence the biotic pressures exerted on fungal interactions by driving changes in the wider soil community [14,19]. As a component of the first trophic level in the decomposer food web, soil fungi support a vast abundance of invertebrates [20]. Although soil invertebrate grazers are known to limit the growth of individual mycelial systems [16,21] and, by feeding selectively on different species, alter the relative abundance of co-existing litter fungi [22,23], only recently has selective grazing strong enough to reverse fungal dominance been revealed [17,24]. Invertebrate macrofauna, particularly woodlice, exert the most consistent influences on individual and interacting mycelia [24,25]. The potential for mycophagous collembola (mesofauna) to influence fungal growth and competitive dominance is determined by grazing intensity [17,[26][27][28].
The influences of elevated temperature and grazing by a range of soil invertebrate taxa on the outcomes of two-species interactions between three common species of temperate woodland decomposer fungi (Hypholoma fasciculare, Phanerochaete velutina and Resinicium bicolor) have been determined (Table 1). The capacity of grazing to alter fungal dominance and responses to abiotic change has implications for top-down regulation of saprotrophic basidiomycete community development. It is, however, unclear whether the fungal dominance hierarchy, and roles of biotic and abiotic factors, apparent from two-species interactions would be sustained or altered (potentially becoming weaker or more idiosyncratic) in more complex assemblages. This study explored the interactive effects of soil invertebrate (collembola or woodlice) grazing and warming on the outcomes of three-species interspecific mycelial interactions and fungal-mediated wood decomposition.
A microcosm approach was employed to overcome limitations to mechanistic understanding of species interactions imposed by the complex and opaque nature of the soil decomposer system [29,30]. The experimental system is a good model of cord-forming basidiomycete ramification at the soil-litter interface, where their mycelia come into direct contact with invertebrate fauna inhabiting soil humus and litter layers. All fungus and invertebrate species used in the study are common in temperate woodland soil. Communities containing these fungal species develop in temperate woodland, where they have been isolated from the decaying wood and litter of several tree species; over time they compete and replace one another within these resources [31,32]. The study aimed to determine: (I) the influence of elevated temperature on fungal dominance and wood decay in three-species interactions; and (II) the extent to which grazing invertebrates moderate the influence of temperature on interaction outcomes and decomposition. Three specific hypotheses were tested: (I) elevated temperature will alter the fungal dominance hierarchy in ungrazed controls; (II) invertebrate grazing will alter competitive interaction outcomes and moderate the influence of elevated temperature; and (III) mycelial growth (measured as extension rate) and wood decay by the dominant competitor will be stimulated by warming.

Experimental design
The influence of soil invertebrate grazing (Folsomia candida and Protophorura armata, Collembola; Oniscus asellus, Isopoda) on the outcomes of three-way competitive interactions between cordforming basidiomycete fungi (Hypholoma fasciculare, Phanerochaete velutina and Resinicium bicolor), at ambient and elevated (ambient +4 uC) temperature, was investigated in compressed soil microcosms. The temperature elevation is within the rise of 1.1 -6.4 uC predicted for temperate regions by 2100 [33]. Organisms living within soil are not expected to experience the same extent of climate warming as will occur in the atmosphere. The soil-litter interface (where the interactions considered in the present study occur), however, will provide little buffering against atmospheric temperature change [34]. Ambient temperature (16 uC) represented late summer -autumn temperatures beneath the litter layer in UK temperate woodland [35]. A fully-factorial experimental design including ungrazed controls was employed, with 5 -6 replicates for all treatments.

Microcosm preparation
Topsoil from mixed deciduous woodland (Tintern, UK, NGR 352800 201800; 51.711, 22.684; permission issued by the UK Forestry Commission) was sieved through a 10 mm mesh on site and air-dried in the laboratory. Dry soil was then sieved through a 2 mm mesh, frozen for 24 h (to prevent population explosions of endogenous collembola; [16]) and mixed with de-ionised water (DH 2 O; 340 ml kg 21 ). Microcosms consisted of lidded clear plastic dishes (24 6 24 cm, 2 cm deep; Nunc-Gibco, Paisley, UK) containing 200 g of soil, which was evenly compressed to ensure fungi grew on the surface of the soil. This enabled visual quantification of mycelial growth and interpretation of interaction outcomes [17,18,24]. Freshly-felled beech (Fagus sylvatica) wood was cut into blocks (2 6 2 6 1 cm) and autoclaved three times at 24 h intervals. Sterile wood blocks were placed on 2% malt agar pre-colonised by the experimental fungi and incubated for 3 months in darkness at 1661uC.
Collembola populations (Cardiff University Culture Collection) were maintained on a medium consisting of 95% plaster of Paris (Minerva Dental, Cardiff, UK) and 5% activated charcoal (Sigma, Poole, UK), in darkness at 20 6 1 uC. Collembola were fed dried baker's yeast (Saccharomyces; Spice of Life, Cardiff, UK), and plaster was re-moistened with DH 2 O, weekly. Prior to experimental use, collembola were size-selected (200-400 mm body width) using a series of stacked sieves (Nickel-Electro Ltd, Weston-super-Mare, UK). Adult O. asellus ($ 5 mm body length) were collected from Cooper's Field, Cardiff, UK (NGR 317819 176785; 51.487, 23.185; permission issued by Cardiff Council). Invertebrates were deprived of food for 24 h prior to addition to microcosms.
Colonised wood blocks were placed 8 cm apart in an equilateral triangular arrangement. Inoculation was staggered (order: R. bicolor, H. fasciculare, then P. velutina) according to species-specific mycelial emergence and extension rates, to ensure mycelia were of equal size when they met [24,25]. This ensured that any treatment effects were a result of temperature or grazing regime and not the relative sizes (which could influence competitive ability) of interacting mycelia. Microcosms were weighed and sprayed weekly with DH 2 O to replace moisture loss. Once opposing mycelia had met (8 cm diameter) in 50% of microcosms, at each temperature (16 and 20 uC), 60 F. candida, 73 P. armata (to provide equal collembola biomass) or 5 O. asellus were added in grazing treatments. Invertebrate numbers represented low field density estimates [36,37]. Microcosms that were not in the required and Resinicium bicolor (unshaded), at ambient (16 6C) and elevated (20 6C) temperature. Each individual fungus could only lose one territory (that which it originally occupied), but had the potential to gain up to two if it replaced both competitors. Size of shaded areas, rather than total heights of the bars, indicate the magnitude of change. Significant (P,0.05) differences in interaction outcomes between grazing treatments are indicated by different letters. * indicates where the interaction outcome at elevated temperature differs significantly (P,0.05) from ambient temperature for the same grazing treatment. doi:10.1371/journal.pone.0077610.g002 standardised condition prior to grazer addition and experimental monitoring (e.g. one of the mycelia failed to emerge from the wood block) were discarded, leaving 5 replicates for all treatment combinations except grazed systems at ambient temperature, which each had 6.

Data collection and analysis
A fungus was considered to gain territory on soil if it directly killed extra-resource mycelium, or overgrew and halted the growth, of an opponent. To gain territory in wood, a fungus had to reach an opponent's wood block and replace the original inhabitant. Wood blocks that had been reached by an opposing species were cut in half and wood chips (approx. 5 6 2 6 2 mm) removed to determine occupancy. Wood chips were dipped in 5% bleach (to prevent contamination), placed on 2% malt agar and incubated in darkness at 16 6 1 uC. The mycelia that emerged were identified visually and only one species was ever isolated from each wood block. Differences in the frequencies of four possible interaction outcomes (H. fasciculare, P. velutina or R. bicolor 'gain territory', or 'draw' if no fungus gained territory), on soil and in wood blocks, between (I) grazing treatments at each temperature, and (II) temperatures for each grazing treatment, were assessed using binomial (two outcomes) or multinomial (three or four outcomes) logistic regression (Minitab 16).
Digital images were captured (Nikon Coolpix 7500 camera; 39.5 cm height under normal laboratory lighting) 0, 2, 6, 12, 20, 30 and 42 d after grazer addition to microcosms. Mycelial extent of each fungus was measured as the mean length of four lines from the wood block centre to the mycelial margin, through a 60u angle towards the opposing fungi (ImageJ; National Institute of Health, USA). Measurement ceased for a species when the first individual in any treatment reached an opposing wood block. Extension rates (cm d 21 ) of each fungus were analysed using two-way analysis of variance (ANOVA) and planned contrasts (R version 2.15.1; [38]), with grazing treatment and temperature as factors.
The state of decay (density, fresh volume/dry weight; g cm 23 ) of each wood block was determined at the end of the experiment (42 d). Decay rates (g cm 23 d 21 ) were estimated after determining pre-treatment density of additional wood blocks (n = 5) from which the relevant fungal species had emerged, at the relevant temperature. Decay rates were analysed using a three-way ANOVA and planned contrasts (R version 2.15.1; [38]), with fungal species, grazing treatment and temperature as factors.

Mycelial extension
Grazing treatment significantly affected mycelial extension rates of all fungi (Table 2; Fig. 3). Oniscus asellus grazing reduced P. velutina (P = 0.004) extension and reversed that of R. bicolor (P,0.001), enabling H. fasciculare to extend more rapidly and at a significantly (P = 0.030) greater rate than in the F. candida treatment (Fig. 3). Elevated temperature did not significantly (P $ 0.05) affect extension of any fungus overall, but reduced that of P. velutina only when grazed by F. candida (grazer*temperature: Table 2; Fig. 3).

Wood decay rates
Decay rates differed between fungi (Table 3); both P. velutina (P = 0.004) and R. bicolor (P = 0.009) decayed wood more rapidly than H. fasciculare. Although warming did increase decay rates, overall (Table 3), this was driven by the effects on H. fasciculare-(P,0.001) and R. bicolor-(P,0.001) mediated decay (temperature*fungus: Table 3; Fig. 4). In contrast, warming reduced decay by P. velutina (P = 0.043; Fig. 4). Impacts of grazing on fungalmediated decay differed between species and were altered by warming (grazer*temperature*fungus: Table 3; Fig. 4). Grazing influenced H. fasciculareand P. veluatina-mediated decay only at elevated temperature, and R. bicolor-mediated decay only at ambient temperature (Fig. 4). Oniscus asellus grazing prevented warming from stimulating decay by H. fasciculare, whereas all grazers intensified the effect of warming on decay by R. bicolor (Fig. 4). Grazing by P. armata and O. asellus increased decay by P. velutina at elevated temperature and prevented warming-induced reduction in decay, which was evident in control and F. candidagrazed treatments (Fig. 4).

Discussion
A clear hierarchy of competitive dominance was evident at ambient temperature (16 uC) from three-species interactions between wood decomposer fungi (P. velutina . R. bicolor . H. fasciculare). Temperature elevated by 4 uC above ambient altered this hierarchy (supporting Hypothesis I) by stimulating the combative ability of R. bicolor such that it became dominant over P. velutina. The pattern of fungal dominance evident at elevated shaded bars) temperature, in the ungrazed control (C), and Folsomia candida (Fc), Protophorura armata (Pa) and Oniscus asellus (Oa) grazing treatments. Different lowercase and uppercase letters indicate significant (P,0.05) differences between grazing treatments at ambient and elevated temperature, respectively. Significant differences between ambient and elevated temperature for a given grazing treatment are indicated (* P,0.05, ** P,0.01, *** P,0.001). doi:10.1371/journal.pone.0077610.g004 temperature supports previous work considering pair-wise interactions between the same three species in similar experimental systems [24]. This is in contrast to interactions in agar culture, where it is often not possible to predict multispecies interaction outcomes due to intransitive dominance hierarchies in pair-wise interactions (e.g. A . B, B . C, C . A; [13]). By promoting competitive dominance of the preferentially consumed species, warming also enabled grazing invertebrates to alter competitive interaction outcomes (supporting Hypothesis II). Resinicium bicolor is generally the most palatable to invertebrates, including those employed in the present study [21,24]. Temperature elevation and grazing, therefore, exerted contrasting influences on fungal community development.
In contrast to Hypothesis (III) and previous work on two-species interactions [17,24], the shift in competitive dominance at elevated temperature was not reflected in increased mycelial growth. Grazing by O. asellus affected mycelial extension; that by R. bicolor and P. velutina was reduced, enabling H. fasciculare to extend slightly (but not significantly) more rapidly. These results are analogous to research on plant communities, in which maximum community biomass was reached at relatively low species diversity levels [39] and grazing by vertebrate herbivores moderated responses to warming [4]. The marked effects of macrofauna on fungal growth and assemblage development supports the conclusion that they exert stronger grazing pressure on individual and interacting mycelia than smaller mesofauna [24,25], even at low density [28]. The capacity for collembola to alter two-species interaction outcomes has been less consistent [17,24]. In the present study, both collembola species influenced multispecies interaction outcomes, at elevated temperature. The stronger effect of woodlouse grazing on P. velutina (after removing the more palatable R. bicolor) enabled the weakest competitor, H. fasciculare, sometimes to gain territory.
Whilst top-down regulation of aboveground [40,41] and bacterial-dominated belowground [42,43] communities is well established, only recently has strong evidence emerged supporting the importance of this process in regulating the structure and function of fungal-dominated decomposer communities [44]. In the present study, increased capacity for top-down determination of the composition of a simple multispecies fungal community by grazing soil invertebrates is largely due to warming-induced promotion of the preferred fungal species. Elevated temperature may also have increased the size of collembola populations, as has previously been reported in microcosm studies with R. bicolor alone [16] and in two-species interactions [17]. The increased grazing pressure exerted by larger collembola populations [26][27][28] provides a greater potential to influence fungal species composition [17]. Macrofauna, including woodlouse and millipede, populations are also predicted to increase due to climate warming [45]. Mycophagous soil invertebrates have previously been hypothesised to drive, and respond to, changes in fungal community compo-sition [14,19]. The present and previous studies demonstrate that grazing by invertebrates is likely to be an important factor influencing the development, and response to climate change, of saprotrophic cord-forming fungal communities in woodland ecosystems.
Although fungal-mediated decomposition of colonised wood inocula was stimulated by elevated temperature, it was not competitively favoured mycelia that displayed the strongest responses (in contrast to Hypothesis III). Decomposition by H. fasciculare and R. bicolor was stimulated more by elevated temperature where their extra-resource mycelia were at a competitive disadvantage, due to either competitive weakness or selective grazing. The shift in fungal dominance hierarchy due to warming in ungrazed controls did, however, reduce resource decay by P. velutina. Species-specific decomposition rates changed, therefore, depending on mycelial competitive success. This suggests that changing species composition of saprotrophic macrofungi and mycophagous soil fauna could alter nutrient cycling and respiratory CO 2 efflux [9]. For example, cellulolytic enzyme production and resource decay rates by R. bicolor are lower than P. velutina [46]. Grazers generally maintained multispecies assemblages at elevated temperature by preventing the competitive exclusion of a species from wood in the majority of cases. As a consequence, all fungal species continued to contribute to 'community' wide decomposition, regardless of competitive dominance on soil. The overall increase in fungal-mediated decomposition due to warming supports previous work in which grazers removed mycelial biomass, but did not prevent temperature-induced stimulation of resource decay [16], and has implications for the promotion of woodland soil nutrient turnover and CO 2 efflux under climate change scenarios.
Although microcosm experiments provide mechanistic insights into otherwise complex and inaccessible ecological interactions, caution must be exercised in extrapolation to more natural conditions where biotic and abiotic variation are much greater [29,30]. Mechanisms of interaction revealed by a microcosm approach will always demand further evaluation under natural conditions. The potential for grazing soil invertebrates to moderate climate-driven shifts in fungal species assemblages supports the argument that biotic and abiotic interactions are important buffers against community responses to climate change [47,48]. Species losses could, therefore, have major implications for the insurance effect of biodiversity [49,50], with functional group, rather than taxonomic, diversity being most important in soil decomposer systems [15,51,52].